Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jul 1.
Published in final edited form as: J Appl Gerontol. 2019 Feb 28;39(7):702–711. doi: 10.1177/0733464819833565

What predicts health care transitions for older adults following introduction of long-term services and supports?

Karen B Hirschman 1, Mark P Toles 2, Alexandra L Hanlon 3, Liming Huang 4, Mary D Naylor 5
PMCID: PMC6713625  NIHMSID: NIHMS1520864  PMID: 30819004

Abstract

Objective.

To determine predictors of health care transitions (i.e., acute care service use, transfers from lower to higher intensity services) among older adults new to long-term services and supports [LTSS].

Methods.

470 new LTSS recipients followed for 24 months. Multivariable Poisson regression modeling within a generalized estimating equation framework.

Results.

Being male, having multiple chronic conditions, lower self-reported physical health ratings and lower quality of life ratings at baseline were associated with increased risk of health care transitions. Older adults in assisted living facilities and nursing homes experienced decreases in health care transitions over time while LTSS recipients at home had no change in risk. LTSS recipients who had orders to receive therapy, compared to those who did not, had a lower relative risk of transitions over time.

Discussion.

Predictors of future health care transitions support the need for LTSS providers to anticipate and monitor this risk for LTSS recipients.

Keywords: hospitalization, emergency department visit, skilled nursing facility, transition, LTSS

BACKGROUND & SIGNIFICANCE

As the population in the U.S. ages, the number of older adults requiring long-term services and support (LTSS) will increase substantially. More than half of all adults 65 and older will develop functional deficits that will require assistance with activities of daily living as well as other health and social services over time (Congressional Budget Office, 2013; Favreault & Day, 2016; Reinhard, Kassner, Houser, & Mollica, 2011). In 2014, nearly seven million U.S. adults, 65 and older, received LTSS assistance in a range of settings, including in their homes via community-based services (HCBS), assisted living communities (ALCs), and nursing homes (NHs) (Nguyen, 2017).

Age-related changes, complicated by frailty and multiple chronic conditions, contribute to high rates of health care transitions among LTSS recipients (Konetzka, Spector, & Limcangco, 2008). In one recent study of over 500,000 Medicare beneficiaries admitted to a NH for either long-term or short-term care, transitions between settings (e.g., hospital to NH, skilled nursing facility [SNF] to NH, NH to home) over a 12 month period occurred among 36.5% of the sample (Li, Middleton, Ottenbacher, & Goodwin, 2017).

Prior epidemiological studies characterize the incidence and predictors of individual transitions in care that predict the use of LTSS and what happens during LTSS use, such as hospital to home, rehabilitation center to home, and skilled nursing facility (SNF) to home (Greiner et al., 2014; Holden et al., 2015; Toles et al., 2014). Findings in observational studies of individual care transitions may under-estimate the risk and burden experienced by older adults who have multiple transitions in care over multiple months or years, often starting prior to LTSS. Thus, determining baseline predictors of multiple health care transitions among those new to LTSS could provide important targets for intervening to optimize health and prevent burdensome transitions when possible, such as, avoidable hospitalizations or emergency department (ED) visits. Moreover, determining risk factors at the start of LTSS is essential for advising recipients and their families about the desired plans of care.

Findings from prior studies reveal common sociodemographic and clinical characteristics that predict hospital admissions, including age, gender, race, socioeconomic status, and specific diagnoses such as heart failure, dementia, and cancer (Callahan et al., 2015; Fabius & Robison, 2017; Kind et al., 2014; O’Connor et al., 2016). Other factors were identified that contribute to differences in rates of hospitalization among this population, including (a) regional practice patterns, (b) quality of care in hospitals, post-acute care settings, NHs and ALCs, hospices, and (c) state Medicaid and federal Medicare NH payment policies (Goodman, Fisher, Chang, & Raymond, 2013; Gruneir, Miller, Intrator, & Mor, 2007; Intrator et al., 2007). Collectively, these factors contribute to frequent use of acute care services and LTSS by frail older adults who often experience negative changes in health status.

Although risk factors for rehospitalization, in general, have been characterized for older adults and older adults receiving LTSS, to the best of our knowledge, no studies to date have examined health care transitions across groups of new LTSS recipients, receiving support from diverse organizations, specifically ALCs, NHs, and HCBS. While findings are available on specific types of health care transitions (primarily hospitalizations), prior research has been limited to specific transitions between settings of care (e.g., hospital to community, SNF/IRF to home; SNF to hospital) (Holden et al., 2015; Toles et al., 2014). Limited attention has been paid to multiple health care transitions among older adults (Li et al., 2017; Wang et al., 2017) but even less is known about multiple health care transitions among ALCs and HCBS recipients. In addition, the relationship between multiple health related quality of life (HRQoL) domains and health care transitions among older adults new to LTSS has not been studied. Without this information, LTSS providers do not optimally anticipate the needs and effectively manage health services for this vulnerable and growing population.

In response to these gaps in knowledge, the purposes of this study were to describe health care transitions over time among a cohort of older adults new to receiving LTSS and to examine the relationships between these health care transitions and individual LTSS recipient sociodemographic and clinical characteristics at baseline.

METHODS

Framework.

The Andersen Behavioral Model of Health Service Use guided this examination of individual and contextual factors contributing to health care use (Andersen, 1995); specifically, the relationship between characteristics of newly enrolled LTSS recipients (e.g., their need, access, and predisposing characteristics) and contextual factors (e.g., setting, health behaviors) that influence heath care transitions. Concepts in the Behavioral Model were defined as follows: (a) health service use was the total number of health care transitions (described in detail below); (b) predisposing factors were the clinical and non-clinical characteristics of new LTSS recipients; (c) enabling factors were characteristics of the health care environment, such as residence in a NH or ALC; and (d) need factors were self-reported health, social support and quality of life of new LTSS recipients.

Data.

This secondary analysis used data from the “Health Related Quality of Life (HRQOL): Elders in Long-Term Care” study (NIA/NINR R01AG025524; Principal Investigator: Mary D. Naylor). The specific aims of the parent study were to describe changes in multiple domains of health related quality life (HRQOL) for a cohort of older adults who at the time of enrollment had just begun to receive LTSS in ALCs, NHs, or in their homes via HCBSs (e.g., Program of All-inclusive Care for the Elderly, Managed Long-Term Care, etc.), and followed for 24 months after enrollment or until withdraw or death. Purposive sampling was used to identifying potentially eligible participants. Individuals were eligible to participate if this was their first experience receiving LTSS (e.g., no previous long-term care service experience), they were within 60 days of the start of services, 60 years of age or older, and willing to complete in-person interviews every three months for two years in English or Spanish, and they had a Mini Mental State Examination (MMSE) score of 12 or higher (Folstein, Folstein, & Fanjiang, 2001; Folstein, Folstein, & McHugh, 1975). Individuals with severe cognitive impairment (e.g., MMSE<12), those with an active impaired sense of reality (such as untreated or poorly managed paranoia or psychosis), or whose plan of care was focused on end-of-life care (e.g., considered terminally ill or enrolled in hospice) were not eligible to participate. According to the parent study, 1311 older adults were referred as potential participants. Among the 1311, 480 (37%) did not meet study eligibility criteria (i.e. >60 days from start of LTSS, n=260; prior LTSS experience, n=144; MMSE<12, n=76). Among the remaining 831 referrals, 327 were screened but declined to participate (n=220), declined in-person visit (n=99) or deferred to enroll and timed out of eligibility (n=8); an additional 29 did not have someone to co-sign and provide additional informed consent on behalf of the participant assessed as having cognitive deficits (MMSE: 12–23) and five were dropped as a result of incomplete enrollment. A total of 470 older adults were enrolled in the longitudinal, observational parent study designed to assess changes in multiple HRQoL domains over time (ALCs, n=156; NHs, n=158; HCBSs, n=156). Enrolled participants came from 11 organizations (59 locations) across three U.S. states (Pennsylvania, New Jersey and New York). Additional details on eligibility and recruitment procedures for the original parent study dataset are described in detail elsewhere (Naylor et al., 2016).

Three Institutional Review Boards (IRBs) – the University of Pennsylvania, the Philadelphia Veterans Medical Center and the Visiting Nurse Service of New York – approved the parent study and were the IRBs of record for the participating organizations. Written informed consent or assent was gathered from all LTSS participants. For each LTSS recipient who wished to participate but who scored 12–23 on the MMSE (Folstein et al., 2001; Folstein et al., 1975), a designated responsible party was asked to provide written informed consent for their family members to participate. At each of the quarterly interviews, participants were reminded of the purpose of the study and asked if they still wished to be interviewed. This secondary data analysis of the parent study dataset was reviewed and approved by the University of Pennsylvania IRB.

Measures.

Counts of health care transitions (primary outcome) and several standardized measures of HRQoL were available from baseline and every three months through 24 months (a maximum of nine data collection time points) for each respondent. Data used for this analysis came from both self-reported data documented by trained research assistants (RAs) at follow-up interviews and chart review abstraction.

Primary Outcome – Total Number of Health Care Transitions.

Similar to recent research on transitions in care (Li et al., 2017; Wang et al., 2017), health care transitions were operationalized as any of the following: (1) acute hospitalizations; (2) emergency department (ED) visits (e.g., transfer to ED that did not result in a hospitalizations); (3) admissions to skilled nursing facilities (SNFs) or inpatient rehabilitation facilities (IRFs); (4) changes in the type of LTSS (e.g., transitions between ALCs to NHs; NHs to ALCs; HCBS to NH); (5) enrollment in hospice; and (6) deaths. Among NH residents who were hospitalized and transitioned back to their NH where they received SNF level care, this was counted as a single transition. For HCSB or ALC residents who had a hospitalization followed by discharge to an IRF before returning home, this counted as two transitions (hospitalization and IRF stay). Data were collected via chart reviews from baseline (3 months prior to the start of LTSS) and then every three months through 24 months from the start of LTSS. The number of events for each type of transition noted above were summed into a count variable in each quarter through the end of the study. The average number of transitions and the total number of transitions in from baseline to each quarter thereafter were calculated.

Baseline Predictor Variables.

In the parent study, individual LTSS recipient characteristics were collected via face-to-face interviews conducted in both English and Spanish by trained RAs and documented on standardized data collection forms; data generated from chart reviews completed by these same RAs at baseline also were documented on such forms. Baseline data included: age, sex, race, ethnicity, years of education, number of chronic conditions, number of medications, nutritional status (Nutrition Screening Initiative Checklist (Posner, Jette, Smith, & Miller, 1993), type of LTSS (i.e., ALCs, NHs or HCBS), ambulation support (e.g., use of cane, walker, wheelchair, brace, etc.) and sensory support (e.g., glasses, hearing aids). Two items from the chart review were used to confirm the LTSS recipient had physicians’ orders to receive physical therapy (yes/no) or occupational therapy (yes/no) at the start of LTSS. Several valid and reliable tools were used to collect self-reported HRQOL domains at study enrollment: (1) overall quality of life [(Godin, Keefe, Kelloway, & Hirdes, 2015)]; (2) cognitive status [MMSE (Folstein et al., 1975)]; (3) symptom status [Symptom Bother Scale (Heidrich & D’Amico, 1993)]; (4) emotional status – depressive symptoms [Geriatric Depression Scale Short Form (Conradsson et al., 2013; Yesavage et al., 1983)] and self-rated mental health [Medical Outcomes Survey Short Form (SF-12) Mental Composite Scores (Ware, Kolinski, Turner-Bowker, & Gandeck, 2002)]; (5) social support [Medical Outcomes Study Social Support Survey (Gomez-Campelo et al., 2014; Sherbourne & Stewart, 1991)]; (6) physical function [Katz Independence in Activities of Daily Living (Katz, Ford, Moskowitz, Jackson, & Jaffe, 1963)]; and (7) general health perception [self-rated physical health, SF-12 physical composite score (Ware et al., 2002)].

Statistical Analyses.

Counts of health care transitions at each data collection point formed the basis for the use of longitudinal multivariable Poisson regression modeling within a generalized estimating equation framework to examine the relationships between health care transitions and potential baseline predictors. The Poisson regression models incorporated an offset to accommodate varying days at risk within each time period. This approach considers censoring for those that drop out of the study. Potential covariates identified in simple main effects and two-way interaction-with-time models were reviewed and included in the building of a final model if the p-value was ≤0.20 (Vittinghoff, Glidden, Shiboski, & McCulloch, 2005). An initial multivariable Poisson regression model was constructed for the primary outcome (i.e., total number of health care transitions) using covariates significantly associated with this outcome and the interaction with time, followed by sequentially eliminating interactions on the basis of least significance until all interaction effects demonstrating significance at the p≤0.20 level remain. Covariates not included in the final interaction model but were significant in the main effect models at the p≤0.20 were then added to the final multivariable interaction model and assessed similarly. The final multivariable Poisson regression model included only interaction terms and covariates demonstrating statistical significance at the p≤0.05 level or based on the literature (i.e., age). Due to known confounding between LTSS type and race (white: ALC, 94%; NH, 33%; HCBS, 27%) and ethnicity (Hispanic: ALC, 2%; NH, 4%; HCBS, 54%) in this dataset (Naylor et al., 2016), neither race nor ethnicity were included in the multivariable modeling. All analyses were performed using SAS 9.4 (SAS Institute, Inc., Cary, NC).

RESULTS

Sample Characteristics at Baseline.

Socio-demographic and clinical characteristics of the 470 LTSS recipients at study enrollment are presented in Table 1. On average, LTSS recipients were 81 years old (range: 60–98); female (71%), widowed (52%) and 85% had five or more chronic health conditions documented in their medical records. Fifty-one percent of the LTSS recipients were White, 34% African American and 12% reported more than one race. Twenty percent of LTSS recipients identified as Hispanic, with 16% of interviews being conducted in Spanish. The majority (64%) of the LTSS recipients had 12 years of education or higher. LTSS recipients were evenly split between the three types of LTSS: NH, n=158; ALC, n=156, HCBS, n=156. Twenty-one percent (10¼07) of LTSS recipients reported being hospitalized in the three months prior to their starting LTSS. At study enrollment, 60% of LTSS recipients had little or no cognitive impairment (MMSE: 24–30), 24% had mild deficits in cognition (MMSE: 19–23), and the remaining 16% had MMSE scores in the moderate range (MMSE12–18). At baseline, LTSS recipients frequently (43%) reported depressive symptoms, indicated by Geriatric Depression Scale scores ≥5, suggesting mild to severe depression (Yesavage et al., 1983). Overall self-rated mental health, as measured with the SF-12 Mental Composite Score, had a mean score of 37, lower than the norm of 50. Overall general health perception, as measured by the SF-12 Physical Composite Score, was 49.0 (norm=50). LTSS recipients reported on average 6 bothersome symptoms at baseline (range: 0–13). Thirty percent reported their overall quality of life “excellent” or “very good”, 33% reported their overall quality of life was “good”, and 35% reported their overall quality of life was “fair” or “poor”.

Table 1.

Sociodemographic and Clinical Characteristics of Older Adults Receiving Long-Term Services and Supports, N=470

Variable N (%) or mean±SD (range)
Characteristics of the Individual
Gender: Female 334 (71.06)
Marital Status
 Married 93 (19.78)
 Widowed 243 (51.70)
 Separated or Divorced 82 (17.45)
 Single (never married) 51(10.85)
 Unknown/Not reported 1 (0.21)
Race
 White 239 (50.85)
 African American 162 (34.47)
 More than one race 57 (12.13)
 Other (Asian/ Native Hawaiian or Pacific Islander/ American Indian or Alaskan Native) 9 (1.91)
 Not Reported 3 (0.64)
Ethnicity, Hispanic 93 (19.79)
Age (years) 80.88±8.71 (60–98)
Education (years) 11.89±4.42 (0–26)
 Missing 1 (0.21)
Number of Chronic Conditions+ 8.63±3.94 (1–27)
 0–4 70 (14.89)
 5 to 8 182 (38.72)
 >8 218 (46.38)
Physical Therapy Orders 294 (62.55)
 Missing 6 (1.27)
Health Related Quality of Life Domains
Symptom Status
 Number of Symptoms 6.1±3.2 (0–13)
  0 to 3 104 (22.13)
  4 to 8 234 (49.79)
  > 8 158 (33.62)
   Missing 18 (3.83)
Functional Status
 Cognitive - Mini Mental State Examination (MMSE) 23.96 ±4.29 (12–30)
  No or little cognitive impairment 278 (59.15)
  Mild cognitive impairment 115 (24.47)
  Moderate cognitive impairment 73 (15.53)
   Missing or unable to complete 4 (0.85)
 Basic Activities of Daily Living 4.3 ±1.9 (0–6)
  Number with no functional deficits 25 (5.32)
  Number with 1–3 functional deficits 108 (22.98)
  Number with 4 or more functional deficits 92 (19.57)
   Missing or unable to complete 17 (3.62)
Self-rated Physical Health
 SF-12 Physical Composite Score 49.01±10.52 (13.89–76.18)
  Missing or unable to complete 17 (3.62)
Emotional Status
 Depressive Symptoms (GDS-SF) 4.55 ±3.39 (0–15)
  0–4 normal 269 (57.23)
  5–10 suggestive of mild to moderate depression 173 (36.81)
  ≥ 10 suggestive of severe depression 53 (11.28)
 Self-rated Mental Health (SF-12 Mental Composite Score) 37.27±10.95 (12.60–61.31)
  Missing or unable to complete 17 (3.62)
Social Support
 Emotional or Informational 2.72 ±1.02 (0–4)
 Tangible 2.97 ±0.95 (0–4)
 Affection 2.79 ±1.18 (0–4)
 Positive Social Interaction 2.46 ±1.14 (0–4)
Overall Quality of Life* 2.97±1.07 (1–5)
 Excellent 40 (8.51)
 Very Good 101 (21.49)
 Good 157 (33.40)
 Fair 127 (27.02)
 Poor 35 (7.45)
 Missing 10 (2.13)
Characteristics of the Environment
 Assisted Living Community 156 (33.19)
 Nursing Home 158 (33.62)
 Home and Community Based 156 (33.19)
Open in a new tab

SD=standard deviation

+

Number of chronic conditions was limited by the clinical documentation system used by one of the H&CBS. An additional review of records and outreach to primary care physicans was conducted to confirm accuracy of number of chronic conditions (80% response rate). Percent may not add to 100 due to rounding.

*

Overall quality of life median=3

Total Number of Health Care Transitions.

Counts by type of health care transition at each quarterly assessment are presented in Table 2. In the three months prior to the start of LTSS, both ALC and NH groups experienced slightly more health care transitions (ALC [mean± standard deviation (SD)]: 0.7±1.0, range: 0–5; NH: 0.7±0.9, range: 0–4) than HCBS (mean±SD: 0.3±0.7, range 0–5). A total of 1294 health care transitions occurred among all LTSS recipients from the three months prior to LTSS through the 24 months follow up period. The most common health care transitions were hospitalizations (654/1195, 55%), followed by ED visits that did not result in a hospitalization (314/1195, 26%) and transfers to SNFs or IRFs for short term rehabilitation (133/1195, 11%). The average number of transitions over the first 12 months were (mean±SD): NH, 2.0±2.5; ALC, 1.4±1.8; HCBS, 1.3±2.0; and greater than 12 to 24 months were NH, 1.3±2.0; ALC, 1.2±1.7; HCBS, 1.3±1.7. Over the full 24 months the group that started in NHs had an average of 3.0 transitions (SD ±3.5); ALCs had an average of 2.2 health care transitions (SD ±2.6) and; HCBS had an average of 2.4 health care transitions (SD ±2.7). Very few people experienced transitions associated with increasing intensity of LTSS (57/1195, 5%). The number of transitions to hospice care also was small, accounting for about 3% (37/1195) of all transitions and only 8% of all LTSS recipients (37/470 recipients). A total of 100 participants died and 34 withdrew (7%) over the 24 months.

Table 2.

Counts of health care transitions and deaths among LTSS recipients over time (months)

Quarter N All health care transitions Hospitalizations ED visits SNF/ IRF LTSS type Hospice enrollment Deaths Missing †
− 3 to 0* months 470 249 133 38 75 0 3 0 0
0* to 3 months 450 236 122 59 20 11 9 15 20
3 to 6 months 439 183 98 43 15 12 6 9 16
6 to 9 months 402 196 85 51 26 8 6 20 44
9 to 12 months 385 160 82 42 14 5 6 11 41
12 to 15 months 368 140 70 35 10 6 7 13 47
15 to 18 months 346 132 66 31 18 6 0 11 56
18 to 21 months 334 124 71 25 16 6 0 6 57
21 to 24 months 327 123 60 28 14 3 3 15 58
Total transitions 0 to 24 months 1294 654 314 133 57 37 100
Open in a new tab
*

= start of LTSS; ED = Emergency Department; SNF/IRF = Skilled nursing facility/ inpatient rehabilitation facilities; LTSS = long-term services- and supports.

†

= reasons for missing data at any given quarter include being unable to make contact with a participant during the interview window timeframe (≤7% of sample), participant refusal to complete a specific follow-up interview (≤4% of sample) or withdraw from study (≤6% of sample).

Multivariate Regression Findings.

The final multivariate Poisson regression model, as depicted in Table 3, revealed that several baseline covariates were associated with health care transitions in this sample. Overall, male LTSS recipients had a 32% higher risk of having health care transitions (male relative risk [RR]: 1.32; 95% Confidence Intervals [CI]: 1.07–1.63; p=0.009). For each additional chronic condition, LTSS recipients had a 3% higher risk of health care transitions (RR: 1.03; 95% CI: 1.00–1.06; p=0.026). LTSS recipients with lower self-rated physical health had a higher risk of health care transitions (for every 10-point decrease in the SF-12 self-rated physical composite score, the risk of any transition increased 10%; RR: 0.90; 95% CI: 0.82–0.99; p=0.03). Finally, LTSS recipients with lower ratings of self-reported overall quality of life had a higher risk of health care transitions (RR: 0.90; 95% CI: 0.81–0.99; p=0.035). The number of transitions at the start of LTSS was highest for those who rated their quality of life as poor (the lowest rating) at baseline compared to those who rated their quality of life as excellent (highest rating); the latter group had the fewest number of transitions.

Table 3.

Multivariable Poisson Regression Model results for number of Transitions over time (baseline to 24 Months by quarter)

Variables RR (95% CI) P-value
Male (Reference group: Female) 1.32 (1.07, 1.63) 0.009
Age (years) 0.99 (0.99, 1.01) 0.281
Number of chronic conditions 1.03 (1.00, 1.06) 0.026
Self-rated Physical Health * 0.90 (0.82, 0.99) 0.030
No orders for PT (Reference group: PT orders) 0.86 (0.59, 1.25) 0.430
LTSS Type (Reference group: Home and Community Based Services; HCBS)
 Assisted Living Community (ALC) 2.06 (1.31, 3.23) 0.002
 Nursing Home (NH) 1.97 (1.38, 2.83) <0.001
Quality of life rating 0.90 (0.81, 0.99) 0.035
Time (3 Month interval) 1.00 (0.98, 1.02) 0.901
Interactions (X) with Time
 No orders for PT X Time (Reference group: PT orders X Time) 1.03 (1.01, 1.06) 0.015
 LTSS X Time (Reference group: HCBS X Time)
  ALC X Time 0.96 (0.93, 0.98) 0.002
  NH X Time 0.98 (0.95, 0.99) 0.036
Open in a new tab

RR=relative risk; CI = confidence interval; LTSS = long-term services and supports; PT = physical therapy

*

expressed in 10 point increments.

Two baseline covariate interactions with time were statistically significant: having physician orders for physical therapy at the start of LTSS (Figure 1a) and initial LTSS type (Figure 1b). LTSS recipients who did not have physicians’ orders for physical therapy at the start of LTSS experience 3% higher risk of health care transitions with each additional month relative to LTSS recipients with a physical therapy order (RR: 1.03; 95% CI: 1.01–1.06; p=0.015). As seen in Figure 1a, LTSS recipients who did not have physicians’ orders for physical therapy at baseline, had fewer transitions during the first 12 months but demonstrated a slight increase in health transitions between months 12 and 24; whereas, those with physicians’ orders for physical therapy at baseline experienced an overall decline in health care transitions over time. Similarly, the baseline LTSS type was significantly associated with health care transitions over time (see Figure 1b). The health care transitions profile among respondents who started in ALCs and NHs differs significantly from those who start LTSS in HCBS. Specifically, those in ALCs experience a 4% lower risk of health care transitions with each additional month relative to those in HCBS. Similarly, participants in NH facilities experience a 2% lower risk of health care transitions with each additional month relative to those in HCBS (RR: 0.96; 95% CI: 0.93–0.98; p=0.002) or NH setting (RR: 0.98; 95% CI: 0.95–0.99; p=0.036).

Figures 1a and 1b.

Figures 1a and 1b.

Open in a new tab

Interactions between Baseline Physical Therapy and LTSS Type over Time

DISCUSSION

As the population of older adults is expected to double in the coming years so too will the number of LTSS recipients, a population with high risk for poor health outcomes resulting in transitions to acute hospitals (Ortman, Velkoff, & Hogan, 2016). In this study, framed with Andersen’s (1995) Behavioral Model of Health Service Use, the total number of health care transitions is described from prior to the start of LTSS through the first two years of LTSS for 470 English- and Spanish-speaking older adults and identified unique participant characteristics associated with health care transitions that predicted changes over time. These findings have several implications.

First, our findings suggest that LTSS recipients commonly experience health care transitions to and from both acute and non-acute settings of care. While much of the health care transitions literature for LTSS recipients has focused on acute hospitalizations (Becker, Boaz, Andel, & DeMuth, 2012; Berish, Applebaum, & Straker, 2016; Cai & Temkin-Greener, 2015; McAndrew, Grabowski, Dangi, & Young, 2016; Sonnega, Robinson, & Levy, 2017), our study also examined non-acute care transitions (e.g., transitions to SNF/IRF settings, to hospice, between LTSS settings) which is a unique contribution to the literature. These non-acute care transitions accounted for 17% of the transitions occurring during the first 24 months of care for LTSS recipients. The prevalence of these transitions suggests the need for transitional care services in areas that are not commonly examined yet have high risk for break downs in communication, poor hand offs and limited care continuity (Becker et al., 2012; Cai & Temkin-Greener, 2015; Sonnega et al., 2017). Additionally, in this prospective study, more health care transitions were observed among older adults in NHs (average 2.0) than in ALCs and HCBSs (averages 1.4 and 1.3, respectively) over the first 12 months period. A similar rate of health care transitions was found in a larger study of older adults who received both short- and long-term NH care (average 2.1 transitions) (Li et al., 2017). We recognize that death is qualitatively different from the other health care transitions, however the final outcome model with and without death changed only slightly and with the inclusion of death in the count resulting in a more clinically meaningful result.

Our findings suggest that, for LTSS recipients in residential settings of care, the number of health care transitions is higher at the start of services and decrease over time while those receiving HCBS have a constant low number of transitions over the two years (see Figure 1b). One implication from this finding is the need for detailed studies of care coordination and nursing care provided by HCBS programs and the impact on function and health care transitions. Moreover, additional examination of health care transitions over time for older adults, including non-acute health care transitions, will provide the basis to adapt transitional care interventions to meet the unique need of LTSS recipients.

Our findings support earlier studies demonstrating individual need, access, and predisposing factors that predict health care use (Becker et al., 2012; Grabowski, Stewart, Broderick, & Coots, 2008; O’Connor et al., 2016; Toles et al., 2014); specifically, important HRQOL domains and characteristics of LTSS recipients influence the number of health care transitions experienced among this population. While available literature describes predictors of hospitalizations (Becker et al., 2012; Grabowski et al., 2008), this is the first study to find that, older adults with physicians’ orders for physical therapy at the start of LTSS, compared to those without physical therapy orders, had fewer health care transitions. This finding suggests the hypothesis that targeting high-risk older adults and implementing preventive strategies, such as physical therapy services, may prevent avoidable health service use. Future research and quality improvement initiatives are necessary to explore this hypothesis. Similarly, LTSS recipients self-reporting higher quality of life at the time of LTSS enrollment was highly associated with the number of health care transitions over time. This finding suggests that using a single-item assessment of overall quality of life can provide valuable information for LTSS providers and identify LTSS recipients with the greatest need for intervention(s) to improve quality of life, such as focusing on maintaining autonomy, dignity, and spiritual well-being (Burack, Weiner, Reinhardt, & Annunziato, 2012).

Our finding that gender was associated with the number of transitions has not been seen in previous literature. There is limited research that supports that older male ALC and NH residents are at high risk for hospitalizations and that male HCBS recipients are at higher risk for transitioning to NHs and needing short term SNF/IRF stays (Becker et al., 2012; Cai & Temkin-Greener, 2015; Grabowski et al., 2008). Further, one study suggests the relationship with gender and long-term care transitions is associated with relationship status and available support in the community (Mudrazija, Thomeer, & Angel, 2015). More research into the relationship with gender and the number of health care transitions among LTSS recipients is needed.

Finally, our finding that lower ratings of general health perceptions and numbers of chronic conditions were significant predictors of the number of health care transitions is supported in the literature; however, participant age and symptoms were not statistically significant in this study, even though they found to be influential in the transition to LTSS and use of health care resources in other studies (Becker et al., 2012; Grabowski et al., 2008; Menec, Nowicki, Blandford, & Veselyuk, 2009; Sonnega et al., 2017). Sensitivity analyses were conducted with the most common transition, hospitalizations, and no symptoms were statistically significant predictors of hospitalizations.

Limitations.

This study has several limitations which we attempted to mitigate. Both self-report and chart review data were used to capture the outcome. While there are limitations to using solely self-report or chart review data, our approach using both types allowed us to capture as much data pertaining to a range of transitions in care as possible. The dataset consisted of ALC, NH and HCBS organizations on the east coast of the U.S. and a more geographically diverse, nationally representative sample may yield different results. The primary outcome, the count of health care transitions, counted each transition as a single event, which is consistent with other literature on hospitalizations and ED visits (Intrator et al., 2007; McAndrew et al., 2016; Wang et al., 2017). For example, a transfer from the ALC to the ED that resulted in a hospital admission, followed by the return to the ALC, was counted as a single health care transition. In addition, the duration (e.g., one hour, one week, one month) and intensity of a health care transition were not captured in the dataset and, thus, a loss of granularity in the impact of these transitions on this population. Future research should capture these detailed data which would add to the growing body of literature on multiple health care transitions as well as provide the basis for understanding individual health care transition patterns for this population. As noted in the methods, race and ethnicity were confounded with LTSS type in this dataset. It is possible that differences by LTSS type may be highlighting unique differences by race and/or ethnicity and socioeconomic status. For example, the ALC group was 94% white and 2% Hispanic, and the NH group was 33% white and 4% Hispanic whereas within the HCBS group there was a great mix of race and ethnicity (27% white, 54% Hispanic). Therefore, subgroup comparisons were not possible due to the complete imbalance between race and ethnicity by LTSS type. Future prospective studies should make a concerted effort to balance on race and ethnicity in order to compare health care transition by LTSS type. Finally, the variable for chronic conditions was a total count of health conditions and not organized by specific conditions. Future research looking at specific conditions such as heart failure, diabetes, dementia should be conducted to determine if there are any unique or common transitions patterns.

Conclusion.

Our findings on the importance of gender, quality of life, number of chronic conditions, and self-rated physical health at the start of LTSS as predictors of future transitions in health care, support the need for LTSS providers to be especially vigilant in anticipating risk for LTSS recipients. By incorporating an assessment of self-rated physical health and quality of life in discussions about goals and preferences for care and in communication with primary care clinicians at the start of LTSS, potentially fewer health care transitions may be seen. As the population of older adults receiving LTSS services continues to grow, the need to understand the nature of health care transitions will also increase. These findings support the need to use a broader definition of ‘health care transition’ to achieve more person-centered, effective and efficient allocation of resources. In addition, the findings suggest next steps for minimizing the risk of potentially preventable transitions, such as evaluating the need for physical therapy consultations for newly admitted LTSS recipients. Finally, additional longitudinal studies are warranted to identify strategies to enhance transitions in care among LTSS recipients.

Acknowledgments:

We would like to thank Bridget Daly for her assistance with initial statistical analyses. We also would like to thank all the participants in this study and the multiple organizations that assisted in recruiting the study sample.

Funding: This work was supported by a pilot grant from the NewCourtland Center for Transitions and Health at the University of Pennsylvania School of Nursing. The dataset used in these analyses was supported by a co-sponsored grant from National Institutes of Health – National Institute on Aging and the National Institute on Nursing Research (R01AG025524), as well as the Marian S. Ware Alzheimer Program at the University of Pennsylvania.

IRB Protocol Numbers: University of Pennsylvania (Protocol # 805326); Visiting Nurse Services of New York (Protocol #: 212462-12); Philadelphia VA Medical Center (Protocol # 01033).

Footnotes

Conflicts of Interest: The authors have no conflicts of interest.

Contributor Information

Karen B. Hirschman, University of Pennsylvania School of Nursing; hirschk@nursing.upenn.edu.

Mark P. Toles, University of North Carolina – Chapel Hill; mtoles@email.unc.edu.

Alexandra L. Hanlon, University of Pennsylvania School of Nursing; alhanlon@nursing.upenn.edu.

Liming Huang, University of Pennsylvania School of Nursing; limingh@nursing.upenn.edu.

Mary D. Naylor, University of Pennsylvania School of Nursing; naylor@nursing.upenn.edu.

References

  1. Andersen RM (1995). Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav, 36(1), 1–10. [PubMed] [Google Scholar]
  2. Becker M, Boaz T, Andel R, & DeMuth A (2012). Predictors of Avoidable Hospitalizations Among Assisted Living Residents. Journal of the American Medical Directors Association, 13(4), 355–359. doi: 10.1016/j.jamda.2011.02.001 [DOI] [PubMed] [Google Scholar]
  3. Berish DE, Applebaum R, & Straker JK (2016). The Residential Long-Term Care Role in Health Care Transitions. Journal of Applied Gerontology, (epub), 1–18. doi: 10.1177/0733464816677188 [DOI] [PubMed] [Google Scholar]
  4. Burack OR, Weiner AS, Reinhardt JP, & Annunziato RA (2012). What Matters Most to Nursing Home Elders: Quality of Life in the Nursing Home. Journal of the American Medical Directors Association, 13(1), 48–53. doi: 10.1016/j.jamda.2010.08.002 [DOI] [PubMed] [Google Scholar]
  5. Cai X, & Temkin-Greener H (2015). Nursing Home Admissions Among Medicaid HCBS Enrollees: Evidence of Racial/Ethnic Disparities or Differences? Med Care, 53(7), 566–573. doi: 10.1097/MLR.0000000000000379 [DOI] [PubMed] [Google Scholar]
  6. Callahan CM, Tu W, Unroe KT, LaMantia MA, Stump TE, & Clark DO (2015). Transitions in Care in a Nationally Representative Sample of Older Americans with Dementia. Journal of the American Geriatrics Society, 63(8), 1495–1502. doi:doi: 10.1111/jgs.13540 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Congressional Budget Office. (2013). Rising Demand for Long-Term Services and Supports for Elderly People Retrieved from http://www.cbo.gov/sites/default/files/cbofiles/attachments/44363-LTC.pdf
  8. Conradsson M, Rosendahl E, Littbrand H, Gustafson Y, Olofsson B, & Lövheim H (2013). Usefulness of the Geriatric Depression Scale 15-item version among very old people with and without cognitive impairment. Aging & Mental Health, 17(5), 638–645. doi: 10.1080/13607863.2012.758231 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Fabius CD, & Robison J (2017). Differences in Living Arrangements Among Older Adults Transitioning Into the Community: Examining the Impact of Race and Choice. Journal of Applied Gerontology, (e-pub), 1–26. doi: 10.1177/0733464816687496 [DOI] [PubMed] [Google Scholar]
  10. Favreault M, & Day J (2016). Long-term Services and Supports for Older Americans: Risks and Financing Retrieved from Washington, DC: https://aspe.hhs.gov/system/files/pdf/106211/ElderLTCrb-rev.pdf [Google Scholar]
  11. Folstein MF, Folstein SE, & Fanjiang G (2001). Mini-Mental State Examination Clinical Guide Lutz, FL: Psychological Assessment Resources, Inc. [Google Scholar]
  12. Folstein MF, Folstein SE, & McHugh PR (1975). “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198. [DOI] [PubMed] [Google Scholar]
  13. Godin J, Keefe J, Kelloway EK, & Hirdes J (2015). Nursing home resident quality of life: testing for measurement equivalence across resident, family, and staff perspectives. Quality of Life Research, 1–10. doi: 10.1007/s11136-015-0989-4 [DOI] [PubMed] [Google Scholar]
  14. Gomez-Campelo P, Perez-Moreno EM, de Burgos-Lunar C, Bragado-Alvarez C, Jimenez-Garcia R, Salinero-Fort MA, … Immigration G (2014). Psychometric properties of the eight-item modified Medical Outcomes Study Social Support Survey based on Spanish outpatients. Quality of Life Research, 23(7), 2073–2078. doi: 10.1007/s11136-014-0651-6 [DOI] [PubMed] [Google Scholar]
  15. Goodman D, Fisher E, Chang C-H, & Raymond S (2013). The Revolving Door: A Report on U.S. Hospital Readmissions Retrieved from Princeton, NJ: http://www.rwjf.org/content/dam/farm/reports/reports/2013/rwjf404178 [Google Scholar]
  16. Grabowski DC, Stewart KA, Broderick SM, & Coots LA (2008). Predictors of nursing home hospitalization: a review of the literature. Med Care Res Rev, 65(1), 3–39. doi: 10.1177/1077558707308754 [DOI] [PubMed] [Google Scholar]
  17. Greiner MA, Qualls LG, Iwata I, White HK, Molony SL, Sullivan MT, … Setoguchi S (2014). Predicting nursing home placement among home- and community-based services program participants. Am J Manag Care, 20(12), e535–536. [PubMed] [Google Scholar]
  18. Gruneir A, Miller SC, Intrator O, & Mor V (2007). Hospitalization of nursing home residents with cognitive impairments: the influence of organizational features and state policies. Gerontologist, 47(4), 447–456. [DOI] [PubMed] [Google Scholar]
  19. Heidrich SM, & D’Amico D (1993). Physical and mental health relationships in the very old. J Community Health Nurs, 10(1), 11–21. [DOI] [PubMed] [Google Scholar]
  20. Holden TR, Smith MA, Bartels CM, Campbell TC, Yu M, & Kind AJ (2015). Hospice Enrollment, Local Hospice Utilization Patterns, and Rehospitalization in Medicare Patients. J Palliat Med, 18(7), 601–612. doi: 10.1089/jpm.2014.0395 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Intrator O, Grabowski DC, Zinn J, Schleinitz M, Feng Z, Miller S, & Mor V (2007). Hospitalization of nursing home residents: the effects of states’ Medicaid payment and bed-hold policies. Health Serv Res, 42(4), 1651–1671. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Katz S, Ford AB, Moskowitz RW, Jackson BA, & Jaffe MW (1963). Studies of Illness in the Aged. The Index of Adl: A Standardized Measure of Biological and Psychosocial Function. JAMA, 185, 914–919. [DOI] [PubMed] [Google Scholar]
  23. Kind AJ, Jencks S, Brock J, Yu M, Bartels C, Ehlenbach W, … Smith M (2014). Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med, 161(11), 765–774. doi: 10.7326/M13-2946 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Konetzka RT, Spector W, & Limcangco MR (2008). Reducing hospitalizations from long-term care settings. Med Care Res Rev, 65(1), 40–66. [DOI] [PubMed] [Google Scholar]
  25. Li S, Middleton A, Ottenbacher KJ, & Goodwin JS (2017). Trajectories Over the First Year of Long-Term Care Nursing Home Residence. J Am Med Dir Assoc doi: 10.1016/j.jamda.2017.09.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. McAndrew RM, Grabowski DC, Dangi A, & Young GJ (2016). Prevalence and patterns of potentially avoidable hospitalizations in the US long-term care setting. Int J Qual Health Care, 28(1), 104–109. doi: 10.1093/intqhc/mzv110 [DOI] [PubMed] [Google Scholar]
  27. Menec VH, Nowicki S, Blandford A, & Veselyuk D (2009). Hospitalizations at the End of Life Among Long-Term Care Residents. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 64A(3), 395–402. doi: 10.1093/gerona/gln034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mudrazija S, Thomeer MB, & Angel JL (2015). Gender Differences in Institutional Long-Term Care Transitions. Women’s health issues : official publication of the Jacobs Institute of Women’s Health, 25(5), 441–449. doi: 10.1016/j.whi.2015.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Naylor MD, Hirschman KB, Hanlon AL, Abbott KM, Bowles KH, Foust J, … Zubritsky C (2016). Factors Associated With Changes in Perceived Quality of Life Among Elderly Recipients of Long-Term Services and Supports. J Am Med Dir Assoc, 17(1), 44–52. doi: 10.1016/j.jamda.2015.07.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Nguyen V (2017). Long-Term Services and Supports Fact Sheet Retrieved from Washington, DC: http://www.aarp.org/content/dam/aarp/ppi/2017-01/Fact%20Sheet%20Long-Term%20Support%20and%20Services.pdf [Google Scholar]
  31. O’Connor M, Murtaugh CM, Shah S, Barron-Vaya Y, Bowles KH, Peng TR, … Feldman PH (2016). Patient Characteristics Predicting Readmission Among Individuals Hospitalized for Heart Failure. Med Care Res Rev, 73(1), 3–40. doi: 10.1177/1077558715595156 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ortman J, Velkoff V, & Hogan H (2016). An aging nation: the older population in the United States population estimates and projections. Current Population Reports. May 2014 Retrieved from https://www.census.gov/prod/2014pubs/p25-1140.pdf
  33. Posner BM, Jette AM, Smith KW, & Miller DR (1993). Nutrition and health risks in the elderly: the nutrition screening initiative. Am J Public Health, 83(7), 972–978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Reinhard SC, Kassner E, Houser A, & Mollica R (2011). Raising Expectations: A State Scorecard on Long-Term Services and Supports for Older Adults, People with Physical Disabilities, and Family Caregivers Retrieved from http://assets.aarp.org/rgcenter/ppi/ltc/ltss_scorecard.pdf
  35. Sherbourne CD, & Stewart AL (1991). The MOS social support survey. Soc Sci Med, 32(6), 705–714. [DOI] [PubMed] [Google Scholar]
  36. Sonnega A, Robinson K, & Levy H (2017). Home and community-based service and other senior service use: Prevalence and characteristics in a national sample. Home Health Care Services Quarterly, 36(1), 16–28. doi: 10.1080/01621424.2016.1268552 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Toles M, Anderson RA, Massing M, Naylor MD, Jackson E, Peacock-Hinton S, & Colon-Emeric C (2014). Restarting the cycle: incidence and predictors of first acute care use after nursing home discharge. J Am Geriatr Soc, 62(1), 79–85. doi: 10.1111/jgs.12602 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Vittinghoff E, Glidden D, Shiboski S, & McCulloch C (2005). Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models New York, NY: Springer. [Google Scholar]
  39. Wang SY, Aldridge MD, Gross CP, Canavan M, Cherlin E, & Bradley E (2017). End-of-Life Care Transition Patterns of Medicare Beneficiaries. J Am Geriatr Soc, 65(7), 1406–1413. doi: 10.1111/jgs.14891 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Ware J, Kolinski M, Turner-Bowker D, & Gandeck B (2002). How to Score Version 2 of the SF-12 Health Survey Retrieved from Lincoln, RI: [Google Scholar]
  41. Yesavage J, Brink T, TL R, Lum O, Huang V, Adey M, & Leirer V (1983). Development and validation of a geriatric depression screening scale: a preliminary report. Journal of Psychiatric Research, 17(1), 37–49. [DOI] [PubMed] [Google Scholar]

RESOURCES